Several readers of this series of posts have themselves posted very interesting comments. Since I am writing this as I post, I have really not had time to respond to these comments as I ought. Be patient, if you will, and I will try to offer some sort of response in a day or two. Now, let me continue.
When we expand our attention to large, complex, modern societies, with hundreds of millions of citizens, the problems attendant on devising usable concepts for grasping so complex a social reality metastasize. The most obvious problem is simply data collection, although that turns out to be much less theoretically significant than one might suppose. When my wife and I go to Paris, and sit in our local cafe, Le Metro, we see many young couples pushing baby carriages. But this sort of anecdotal evidence is obviously useless in telling me whether the birth rate in France is above or below the replacement rate. [It is, as it happens, below -- without immigrants, most of whom are reviled by the French, the population would be declining precipitously]. To form a usable concept of the rate of expansion or contraction of the population of France, one needs population statistics.
In the United States, statistics of all sorts are regularly collected by the Bureau of Labor Statistics, an arm of the Labor Department of the Federal Government. If this subject interests you at all, I strongly urge you to familiarize yourself with their work. [Any University Library that is a "repository library" will have the full run of their publications. At one point, when I was deep in the details of employment and compensation statistics by job category, I had occasion to email the Bureau, and ended up in contact with a fascinating character who was quite comfortable with my Marxist beliefs. He was spending his career in some obscure government office, possessed of radical leanings that rarely found an avenue of expression.]
There are two theoretical problems with the social statistics collected by the Bureau and used by virtually everyone in getting a conceptual grasp of American society. In brief, these are the sampling problem and the index number problem. The first is manageable although it is a good deal more complex than one might at first imagine. The second is completely insoluble, a fact that has the most profound implications for our understanding of social reality.
Every month, the BLS [i.e., Bureau of Labor Statistics] tells us how many new jobs have been created. How on earth do they know that? A little thought will convince you that they cannot possibly count every man and woman hired by every employer in America. The answer, of course, is that they carry out a carefully constructed sampling, and then extrapolate. [Literature tip: If you have the time, take a look at the fine old book by Oskar Morgenstern called ON THE ACCURACY OF ECONOMIC OBSERVATIONS -- this is the same Morgenstern who is the junior author of one of the great books of the twentieth century, von Neumann and Morgenstern's THE THEORY OF GAMES AND ECONOMIC BEHAVIOR.] Every month, the BLs defines a sample of employers [and another for households, etc.], constructed so as -- they hope -- to mirror the array of employers nation-wide. Then they ask the employers in that sample a number of questions: How many new workers did you employ last month? How many of those were full time? What wages did you pay them? How many employees do you anticipate hiring -- or firing -- in the coming month? And so forth. Then they extrapolate to national figures.
The technical problems with designing sample and then calculating the probability that it successfully mirrors the total population of which it is sample are well known and have for more than a century been the subject of sophisticated theoretical investigation. [I am going to assume that the readers of this series of blog posts are sophisticated about such matters, and understand concepts like "margin of error" and "standard deviation." If not, perhaps I will try to say something about them.] But there are problems lurking even in this region. Let me take a moment to talk about just one such problem -- the definition of a national unemployment rate.
These days, unemployment statistics are constantly in the news, as more and more Americans go for extended periods of time without being able to find jobs. At first glance, the determination of the unemployment rate seems to be a straightforward sampling problem, but a closer examination reveals very deep difficulties that will bring us closer to the thesis I am leading up to asserting. [I should indicate that the problems I am about to discuss are well known by the folks at the BLS, who are really quite intelligent and sophisticated economists. They are constantly struggling to deal with these problems, in ways that deepen our understanding of American society.]
What is an "unemployment rate?" It is presumably the proportion of the labor force that does not have a full-time paying job. So to say that the unemployment rate is 9.5% is to say that 95 out of every one thousand persons in the labor force are without full-time paid employment. We imagine a city [or state or nation] in which a certain group of men [and women -- this raises interesting problems], the labor force, wake up each morning to go to work, but 9.5% percent of them have no full-time jobs to go to. So the unemployment rate is 9.5%. [I keep saying "full time employment" because there may be a great many people employed part-time, some of them by choice, some of them of necessity.]
But just who is in the labor force? Are all Americans above a certain age [say, sixteen] in the labor force? Are all Americans sixteen or over who are not in school in the labor force? Are all women over the age of sixteen in the labor force, even those who conceive of themselves as housewives, whether or not they have small children at home? Is a woman with a ten year old in the labor force even if she believes that it is morally incumbent upon her to stay at home and make dinner for her husband and be there for her child during all of the hours when she doesn't have school? Is the prison population of the United States in the labor force? Are members of the Armed Forces in the labor force, even though it is not open to them to leave that job, if it is a job, and look for another one?
These are not frivolous questions. From the perspective of Capitalism, which cares only about market relations and profits, housewives are not in the labor force unless they look for paying jobs. But from the point of view of some Scandinavian countries, housewives, because they perform a vital productive function, are part of the labor force. In those countries, household workers are included in the labor force figures, and elaborate calculations are made of the wages implicitly earned by housewives performing essential productive functions. [If you doubt that housewives do perform essential productive functions, ask yourself how long a society will continue to exist if no one raises the next generation of paid workers -- indeed, how long the society will exist if no one cooks diner for the paid workers and does their laundry and maintains the household in which they live when they are not at work.]
How does the BLS handle these questions? Tomorrow, I will talk about that, and the problems that are posed by any attempt to define the labor force.